Deep Learning Approach for Automatic Classification of Ocular and Cardiac Artifacts in MEG Data

Figure 7

Receiver operator characteristics (ROC) are shown separately for the spatial and temporal branch and the combined model as implemented in DCNN. The ROC curves show the sensitivity (true positive rate, TPR) against the fallout (false positive rate, FPR) for various thresholds. (a) Total view on the ROC curves in comparison to a totally random classification (dashed line). (b) The close-up of the ROC curves highlights the differences in model performance.

(a)

(b)